Supply chain, finance, medical, aerospace, defense, cybersecurity, GPU clusters, and LLM evaluation — if it's a table with outcomes, Nostrix Math ranks it, learns from it, and explains it.
Any CSV works. Tag headers with Revenue:V+, Cost:H- — the engine handles direction, numbered dims, and priority weights automatically.
Level 1 learns 4 dimension weights (Q/V/G/H). Level 2 learns a weight for every metric. Elastic Net with L1 zeroes worthless features automatically.
Re-fit weights as real-world behavior shifts. Identify the single most influential metric and act on it. The unsolved MLOps problem, made explainable.
Taylor-series prediction with velocity, acceleration, and jerk per metric. See where a sequence is heading, not just where it is.
Scores 500 tickers and runs the full learning pipeline in a blazing native binary — built for fleet-scale telemetry and node ranking.
A full terminal workflow plus an enterprise JSON API and a spreadsheet-style web grid. Score from any pipe, any language.
Layer 1 scores dimensions. Layer 2 computes cross-sectional derivatives. Layer 3 outputs a decision you can act on.
Quality, Value, Growth, History. Weighted from defaults, P-tags, or learned from your outcomes.
Mean deviation, spread, quality-growth gap, and compound interaction per entity. The math that distinguishes similar rows.
A final 0–100 score, probability, Kelly fraction, risk, and upside potential. One clear verdict per entity.